What chart patterns are, and what they are not
A chart pattern is a shape that price traces over a stretch of candles: a pair of peaks, a narrowing wedge, a sharp move followed by a tight pause. Traders have named these shapes for over a century because the same human behavior keeps producing similar-looking charts. Buyers get exhausted, sellers get trapped, volatility contracts and then expands.
What a pattern is not: a prediction. It is a compact description of a fight between buyers and sellers, plus a rule of thumb about which side tends to win after the fight ends. The rule of thumb fails often enough that every pattern needs a stop-loss attached, which is covered in stop-loss and take-profit explained.
Reversal vs continuation patterns
Every pattern fits one of two families, and the family tells you what the shape is arguing.
Reversal patterns appear after a trend has run for a while. They argue that the dominant side is tiring. Head and shoulders, double tops and double bottoms are the classic examples.
Continuation patterns appear in the middle of a trend. They argue that the move paused to catch its breath and will probably resume. Flags, pennants and most triangles belong here.
The word "probably" matters. A triangle in an uptrend is a continuation pattern by convention, yet plenty of them break down. The family tells you which breakout direction the textbook expects, not which one you will get.
Head and shoulders
A head and shoulders top has three peaks. The middle peak (the head) is higher than the two outer peaks (the shoulders), and the lows between the peaks line up roughly along a line called the neckline. The story: buyers pushed to a new high, failed to repeat it, and each rally since has been weaker.
The pattern is only considered complete when price closes below the neckline. Before that, it is just three peaks. The inverse version, an inverse head and shoulders, appears after a downtrend and signals a possible bottom when price closes above the neckline.
Two cautions. First, the shoulders are never perfectly symmetrical, so there is a lot of room for wishful thinking. Second, the pattern needs a prior trend to reverse. Three bumps in a flat range are not a head and shoulders.
Double top and double bottom
A double top is two peaks at roughly the same price with a trough between them. The trough low acts as the neckline. Price failing twice at the same zone suggests supply is sitting there, the same idea behind support and resistance. The pattern completes on a close below the trough.
A double bottom is the mirror image: two lows at a similar level, confirmed by a close above the peak between them.
The common error is calling a double top after the second peak forms. At that moment you only have two peaks. Price can easily push through for a third, and then the "double top" was just resistance that eventually gave way.
Flags and pennants
A bull flag is a sharp rally (the pole) followed by a shallow, orderly pullback that slopes slightly against the trend (the flag). A pennant is the same idea, but the pause converges into a small symmetrical triangle instead of a tilted channel. Bear flags are the downward mirror.
What makes a flag a flag: the pole is fast, the pause is short and contained, and volume typically fades during the pause. A pullback that retraces most of the pole is no longer a flag, it is a failed move.
Flags are popular because the stop is easy to place: below the low of the pause for a bull flag. That tight, defined risk is the real appeal, more than any promise about what comes next.
Triangles
Triangles are consolidations where highs and lows squeeze toward each other.
- Ascending triangle: a flat top with rising lows. Buyers keep stepping in higher while sellers defend one level. By convention, a bullish breakout is favored.
- Descending triangle: a flat bottom with falling highs. Sellers keep pressing lower while buyers defend one level. A bearish breakdown is favored.
- Symmetrical triangle: both boundaries converge, with no flat side. It is neutral until it breaks, and the break can go either way.
Think of it as volatility compressing. Compression usually ends in expansion. The pattern says that a bigger move is coming; it is far less reliable about direction.
Wedges
A wedge also narrows, but both boundaries slope in the same direction. A rising wedge climbs on shrinking momentum, and is often read as bearish. A falling wedge drifts down on fading selling pressure, and is often read as bullish.
Wedges are the easiest pattern to over-draw. Almost any short uptrend with two tilted lines can be called a rising wedge afterwards. A momentum check helps keep you honest: if RSI is making lower highs while price makes higher highs inside the wedge, the "weakening" claim has some evidence behind it. Without that, it is mostly a drawing.
How breakouts are confirmed
Every pattern above resolves at a boundary: a neckline, a flag edge, a triangle line. The question is how to tell a real break from a poke.
- Close beyond the level. A candle that closes outside the boundary says more than a wick that touched it. Wicks beyond a level are exactly what stop hunts produce.
- Volume. A breakout that comes with volume above the recent average has more participation behind it. Low-volume breaks are more likely to fade.
- A retest. Price often returns to the broken boundary. If the old resistance holds as new support, the break has passed a second test.
- Context. A breakout into a nearby major level, or against the higher-timeframe trend, deserves more suspicion.
You can stack all of these and still get a false breakout. Confirmation reduces how often you are fooled and increases how often you are late. That is the trade-off, and there is no free version.
Measured-move targets are a convention, not a promise
The usual way to set a target is the measured move: take the height of the pattern and project it from the breakout point. A double top whose peaks sit 10% above the trough projects a target roughly 10% below the trough.
This is useful because it gives you a number to compare against your stop, which tells you whether the trade offers enough reward for the risk. It is not useful as a forecast. Price routinely stalls halfway, overshoots, or reverses straight back through the breakout. Use the projection to plan a take-profit and size the trade, not to believe in a destination.
A worked example (hypothetical numbers)
This is an illustration with invented prices, not a real trade.
Suppose a coin rallies from 80 to 100, then forms a tight pause between 94 and 100 over several candles on falling volume. That is a bull flag: pole of 20, flag between 94 and 100.
A trader waits for a candle to close above 100 with volume above the recent average. They enter at 101, put the stop just below the flag low at 93, so the risk is 8 per unit. The measured-move target is the pole height added to the breakout: 100 + 20 = 120, so the potential reward is 19 per unit against 8 risked, a ratio of about 2.4 to 1.
Now the honest part. Three outcomes are all normal:
- Price hits 120 and the trade works.
- Price spikes to 101.5, closes back inside the flag, and hits the stop at 93. A false breakout, a loss of 8.
- Price rises to 112, stalls, and falls back. The target was never reached.
A trader using this setup is not betting on any single outcome. They are checking that the loss in case 2 is small enough to survive being wrong several times in a row.
The biggest beginner mistake
Pattern-hunting after the fact. You scroll back through a chart, see a beautiful double top, and conclude the pattern is reliable. Of course it looks great: you picked it knowing how it ended. The ones that did not work are quietly skipped, and the ones that were never drawn because they did not look right at the time never enter your memory.
The live version of the same mistake is calling the pattern early. Beginners buy the "bull flag" before the breakout, short the "double top" before the neckline breaks, and treat an unfinished shape as a confirmed one. Wait for the close, or accept that you are guessing.
A close relative of this mistake is ignoring the regime. A bull flag in a strong uptrend and a bull flag in a choppy range may look identical and behave very differently. In trends, continuation patterns have momentum behind them. In ranges, breakouts tend to be fakeouts that snap back inside. Reading market regimes before you trust a shape is worth more than learning another pattern. Volatility context from Bollinger Bands can help too: a contracting triangle often coincides with narrowing bands.
How systematic traders treat chart patterns
Here is the uncomfortable fact about chart patterns: they are hard to define objectively. How many candles make a flag? How equal must two tops be to count as a double top? Where exactly does the neckline sit when the lows are uneven? Ask ten experienced chartists and you get ten drawings. A rule that cannot be written down precisely cannot be tested precisely, and anything you cannot test honestly is closer to opinion than evidence.
That is why systematic trading engines tend to prefer measurable rules: a close above the highest high of the last N candles, a volatility contraction below a stated threshold, a momentum reading crossing a stated level. These capture part of what pattern traders see, but each one is unambiguous, so it can be tested on forward data instead of argued about.
Darwin Lab is a self-evolving genetic-algorithm bot for Binance Futures. It traded real money from April to June 2026 and has run on paper since June 2026. Its strategies are evolved and promoted on forward paper results, not backtests, and it classifies the market regime before weighting signals. It does not claim to detect chart patterns. It posts its signals publicly, so you can watch the results rather than take anyone's word for them. For the full picture, see How Darwin Lab works.
Nothing here is financial advice. Chart patterns are one lens for reading price; treat them as context, define your risk before you act, and expect a share of them to fail.